Doing knowledge managementJoseph M. Firestone

Executive Information Systems Inc., Alexandria, Virginia, USA, and

Mark W. McElroyCenter for Sustainable Innovation, Windsor, Vermont, USA

Abstract

Purpose – Knowledge management (KM) as a field has been characterized by great confusion aboutits conceptual foundations and scope, much to the detriment of assessments of its impact and trackrecord. The purpose of this paper is to contribute toward defining the scope of KM and ending theconfusion, by presenting a conceptual framework and set of criteria for evaluating whether claimedKM interventions are bona fide instances of it or are interventions of another sort.

Design/methodology/approach – Methods used include conceptual evaluation and critique of avariety of types of “KM interventions” and presentation of a detailed analysis of an unambiguous case(The Partners HealthCare case) where KM has been successful.

Findings – The critical analysis indicates that the use of tools and methods associated with KM doesnot imply that interventions using them are KM interventions, and most “KM projects” are probablyinterventions of other types. The analysis also illustrates a pattern of intervention that can serve as thebasis of a long-term systematic strategy for implementing KM.

Originality/value – This is the first detailed examination of whether KM is really being done bythose who claim to be doing it. It should be of value to all those who think about the scope oforganizational learning and KM, and who care about unbiased assessments of its performance.

Keywords Knowledge management, Strategic management, Problem solving, Quality control

Paper type General review

IntroductionHas knowledge management (KM) been done? Of course, KM has been done. It is anatural function in human organizations, and it is being done all of the time in aninformal distributed way by everyone undertaking activity in order to enhanceknowledge production and integration tasks. But whether formal interventionsclaiming the label “KM” are bona fide instances of KM practice is another matterentirely. To answer that question, we need to have clear, non-contradictory ideas aboutthe nature of knowledge, knowledge processing, and KM. And to have those, we needto get beyond the notion that we can do KM by just doing anything that may have apositive impact on worker effectiveness while calling that thing “KM”.

Instead we need to recognize that the immediate purpose of KM is not to improveeither worker effectiveness (although it may well do that) or an organization’s bottomline. Its purpose is to enhance knowledge processing (Firestone and McElroy, 2003a,Ch. 3) in the expectation that such enhancements will produce better quality solutions(knowledge), which, in turn, may, ceteris paribus, when used, improve workereffectiveness and the bottom line. And when we undertake KM projects, we mustevaluate the contributions of our interventions to the quality of knowledge processingand knowledge outcomes. That calls for tough, precise thinking about knowledgeprocessing, knowledge, and the impact on these that our interventions are likely tohave.

The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at

www.emeraldinsight.com/researchregister www.emeraldinsight.com/0969-6474.htm

Doingknowledge

management

189

The Learning OrganizationVol. 12 No. 2, 2005

pp. 189-212q Emerald Group Publishing Limited

0969-6474DOI 10.1108/09696470510583557

The question we are asking here is whether KM practitioners are, in fact, providingthis tough, precise thinking as a basis for KM practice, or whether, instead, they are“practicing KM” by helping fields or techniques such as information technology (IT),content management (CM), customer relationship management (CRM), social networkanalysis, storytelling, communities of practice (CoPs), and “knowledge” cafes to“colonize” it? Is such conceptual drift in KM so widespread that one can conclude that,generally speaking, at least, KM as a formal, intentional endeavor has, indeed, not yetbeen done?

In this paper we will begin by providing an account of our view of KM, knowledgeprocessing, information, knowledge, and KM, and then continue by considering theabove questions and by analyzing the Partners HealthCare case, a case where KM hasmost emphatically been done, and done successfully. We will then end by drawing outthe implications of the Partners HealthCare case for KM strategy and KM programs.

The nature of KM as a type of activity or a set of processesIn an earlier “Viewpoint” in TLO (Firestone and McElroy, 2004a) we presented athree-tier framework (see Figure 1) of business processes and outcomes (also seeMcElroy, 2003; Firestone, 2003a; Firestone and McElroy, 2003a, b), distinguishingoperational business processes, knowledge processes, and processes for managingknowledge processes. Operational processes are those that use knowledge but, apartfrom routinely produced knowledge about specific events and conditions, do notproduce or integrate it. Examples of outcomes are sales revenue, market share,customer retention and environmental compliance.

There are two knowledge processes: knowledge production, the process anorganization executes that produces new general knowledge; and other knowledgewhose creation is non-routine; and knowledge integration, the process that presents

Figure 1.The three-tier framework

TLO12,2

190

this new knowledge to individuals and groups comprising the organization. Examplesof outcomes are new organizational strategies communicated throughout an enterpriseusing e-mail, and new health insurance policies communicated through a new releaseof the organization’s personnel manual.

KM is the set of processes that seeks to change the organization’s present pattern ofknowledge processing to enhance both it and its outcomes. A discrete KM activity isone that has the same goal as above or that is meant to contribute to that set ofprocesses. The discipline of KM is the study of such processes and their impact onknowledge and operational processing and outcomes. The foregoing implies that KMdoes not directly manage, create or integrate most knowledge outcomes inorganizations, but only impacts knowledge processes (performed by operationalprocess agents), which, in turn, impact knowledge outcomes. For example, if aknowledge manager changes the rules affecting knowledge production, then thequality of knowledge claims may improve. Or if a KM intervention supplies a newsearch technology, based on semantic analysis of knowledge bases, then that mayresult in improvement in the quality of business forecasting models.

The context of KM: CASs, DECs, and learningWhat is the conceptual context of this three-tier conceptualization of KM? It is in theintegration of the theories of complex adaptive systems (CASs) (Holland, 1995;Gell-Mann, 1994; Kauffman, 1995; Juarrero, 1999; Hall, 2005) and organizationallearning (OL) (Argyris and Schon, 1974; Argyris, 1993; Senge, 1990). The three types ofprocesses distinguished in the three-tier framework occur within complex adaptiveorganizational systems that are characterized by distributed continuous learning andproblem solving, self-organizing, and emergent phenomena produced by dynamicprocesses of interacting autonomous agents that are non-deterministic in character(Holland, 1998). Emergent phenomena at the group and global system levels inorganizations exhibit “downward causation” on individual decision makers in suchsystems (Campbell, 1974; Bickhard, 2000). These phenomena include social,geo-physical, economic, and cultural conditions, and also social network effectspresented to individuals in the form of transactions directed at them by other decisionmakers who collectively constitute the emergent network pattern (see Figure 2) of theorganizational CAS (Firestone and McElroy, 2003a, Chs 2 and 4).

When we look more closely at individual CAS agents and their decisions, weconnect to matters that have received a great deal of attention in the field oforganizational learning. Decisions are part of a sequence of cognitive operations thathave been described in the literature in slightly varying terms, using many names (e.g.the organizational leaning cycle (Ackoff, 1970), the experiential learning cycle (Kolband Fry, 1975; Kolb, 1984), the adaptive loop (Haeckel, 1999), and others). We call it theDecision Execution Cycle (DEC), which includes planning, acting, monitoring, andevaluating behaviors (Firestone, 2000a). Decisions are produced by planning and areembodied in acting. Decisions produce actions; and actions – activities – are the stuffthat social processes, social networks, and (complex adaptive) organizational systemsare made of. Figure 3 illustrates the phases of DECs.

DECs use previously existing individual-level knowledge to arrive at decisions andactions. Personal knowledge is always the immediate precursor to action. DECs alsogenerate new knowledge about specific conditions and situations by using preexisting

Doingknowledge

management

191

Figure 2.The organizational CASnetwork

TLO12,2

192

knowledge in a routine way to monitor, evaluate, plan and decide. This is thesingle-loop learning (SLL) of Argyris and Schon (1974). In addition, DECs play a keyrole in initiating and performing double-loop learning (DLL) (Argyris and Schon, 1974)– learning of new knowledge (in the form of general predispositions and rules, andspecific knowledge) that requires problem solving and is not just a matter of perceptionor direct apprehension or comprehension.

Elsewhere, we (Firestone and McElroy, 2003a, c; Firestone, 2003a, c) have describedhow routine DECs give rise to DLL. In brief, DEC decisions and actions areaccompanied by expectations. During monitoring and evaluating, the individualdetermines the degree to which results match the expectations accompanyingdecisions, and when mismatches occur, the seriousness of the mismatch from both thefactual and evaluative perspectives (see Figure 4). When the mismatch is great enoughfrom the viewpoint of the individual, and when the individual decides that previousknowledge will not work to reduce the mismatch, the individual recognizes that a gapexists between what the individual knows and what she or he needs to know in order topursue the goal(s) or objective(s) of the associated DECs. This knowledge, or epistemic,

Figure 3.The decision execution

cycle

Figure 4.The decision execution

cycle and problemrecognition

Doingknowledge

management

193

gap is what we mean by a “problem”, and recognition of it is what we mean by“problem recognition”.

When a DEC results in problem recognition, the individual can either abandon orsuspend pursuing the goal or objective motivating associated DECs or alternatively,the individual can engage in problem solving or DLL, a process composed of multiplelearning-related DECs motivated by a learning incentive. Following Popper (1999), weview DLL most generally as an emergent (i.e. non-deterministic) three-stage knowledgeprocess comprised of problem formulation, developing alternative solutions, and errorelimination, the stage in which we select among alternatives by eliminating the oneswe think are false. Among the results of error elimination is knowledge, which we willdiscuss briefly below. Here we call attention to the need, once new knowledge isproduced, for further knowledge processing to integrate it into the DEC and businessprocess environment that originated it, and into the organizational memory that willmake it available for re-use later.

The knowledge life cycle, the business processing environment, and theDECSo far, our account of DLL/problem solving as involving sequences of DECs hasfocused on the individual level of analysis. However, DECs may also form patterns ofinterpersonal collaboration, cooperation, and conflict, and these patterns may alsointegrate into knowledge processes. When they do, we can differentiate betweenproblem formulation, developing alternative solutions, and error elimination, on theone hand, and problem claim formulation, knowledge claim formulation, andknowledge claim evaluation in order to distinguish the individual level of knowledgeprocessing from the interpersonal and collective levels, respectively. We alsodistinguish information acquisition and individual and group learning, as additionalknowledge sub-processes preceding knowledge claim formulation. Informationacquisition includes activities of finding and retrieving knowledge claims producedin external systems. Individual and group learning is a category identifying levels ofknowledge processing nested within the knowledge production process beinganalyzed. Individual and group learning produces knowledge from the viewpoint ofnested knowledge processes, and knowledge claims from the viewpoint of knowledgeclaim formulation at higher levels of analysis.

When we view knowledge processing at levels of analysis higher than theindividual level, we identify the pattern including problem claim formulation,information acquisition, individual and group learning, knowledge claim formulation,and knowledge claim evaluation as the knowledge production process resulting in bothnew tested and surviving beliefs and knowledge claims. Once new knowledge isproduced at the collective level, it must be integrated into organizational memory, keyDECs and business processes. This process of knowledge integration is made up offour more sub-processes, all of which may use interpersonal, electronic, or both types ofmethods in execution. They are: knowledge and information broadcasting;searching/retrieving; knowledge sharing (peer-to-peer presentation of previouslyproduced knowledge); and teaching (hierarchical presentation of previously producedknowledge).

Knowledge integration is about system-level knowledge claims beingcommunicated from one part of the distributed organizational knowledge base

TLO12,2

194

(DOKB), the configuration of previously produced knowledge claims, beliefs and beliefpredispositions in the organization (Firestone and McElroy, 2003a) (see Figure 5), toanother. Knowledge claims are stored in media and information systems. Beliefs andbelief predispositions are stored in minds. Through the DOKB, both knowledge claimsand belief phenomena are accessible in varying degrees to individual decision makersin DECs, within both the business processing environment, and the knowledge and KMprocessing environments. That is, the DOKB is the knowledge and informationfoundation for all of the organization’s DECs and processing environments. Whenknowledge claims are evaluated, results of evaluation in the form of changes in beliefsand new knowledge claims, including those we call “meta-claims” which provide the“track record” of criticism, testing, and evaluation of knowledge claims producedduring knowledge claim formulation, are stored in the DOKB. Knowledge claims, aswell as meta-claims, are then integrated and reintegrated into the DOKB as they arebroadcasted, retrieved, shared and taught again and again.

A visual of knowledge processing and its relationship to operational businessprocessing is given in Figure 6, the Knowledge Life Cycle (KLC) (McElroy, 1999, 2000,2003; Firestone, 2000b, 2003b, Firestone and McElroy, 2003a, b, c; Cavaleri and Reed,2000, 2001). Actually, the KLC extends from problem claim formulation to theintegration of knowledge and information in the DOKB. Knowledge claim evaluation(KCE) occupies a central place in the visual and in knowledge production. It is KCEthat produces surviving, falsified, and undecided knowledge claims, and alsometa-claims, for storage in the DOKB. Of course, the extent to which this “track record”is stored or lost depends on the specifics of each organization. The bottom of the figureillustrates the workings of the business processing environment, including its role inusing knowledge for business processes and in recognizing problems that arisethrough mismatches of results and expectations, which, in turn, initiateDLL/knowledge production activity.

Figure 5.The distributed

organizational knowledgebase (DOKB)

Doingknowledge

management

195

Figure 6.The knowledge life cycle(KLC) and the businessprocessing environment

TLO12,2

196

The clouds in the figure illustrate the ubiquity of DOKB content in the variousprocesses. We have also used arrows from the primary DOKB cloud to illustrate itsinfluence on all processes, but are limited to showing its universal influence in twodimensions, while at the same time showing the breakdown of primary knowledgeprocesses into sub-processes and other details in the figure.

Since Figure 6 focuses on a process view, it glosses over the lower DEC level ofanalysis. Figure 7 makes it clear that the match/mismatch process occurs in DECs andnot simply at the higher level of business processes. This point is very important forour later analysis of the Partners HealthCare case.

Information and knowledgeInformation is a non-random structure within a system, indicating future interactivepotentialities, either originating along with it, or acquired or developed by it in thecourse of its interacting with and responding to its environment and the problemsgenerated by that interaction (Bickhard, 1999). Note that this definition does notrequire correspondence between information and the environment. Nor does it assertthat information is encoded in some simple cause-and-effect fashion, but leaves roomfor emergent information in the context of interaction with the environment.

The most important aspect of information, in our view, however, is not whether it iscomplex or simple, or produced quickly or slowly, or gained or lost over time, orwhether there is a great or a small amount of it. All of these are undoubtedly important,but the most important aspect of information is whether its influence on behaviorenhances the ability of the system using it to adapt. And this ability to adapt, in turn, ismost likely to be enhanced if the information itself actually corresponds to the reality ofthe system’s environment. Evolution provides such correspondence by selecting forthose life forms that fit the environmental constraints in which they live. Errors ingenetic information are eliminated over time by the environment, along with theorganisms that contain them (Popper, 1987). Learning provides such correspondenceon a much shorter time scale by providing us with an opportunity to eliminate our

Figure 7.Matches, mismatches and

the DEC

Doingknowledge

management

197

errors in information and to create new information that survives our evaluative effortsand our experience.

Since the most important aspect of information is correspondence with reality, themost important measures of information networks are those that evaluate thiscorrespondence. Thus, the most important measures we can develop describingknowledge claim (information) networks are measures that help us to evaluateknowledge claims, and that brings us to “knowledge”. One of the moments of truth inany consideration of KM is when it is time to say what one means by “knowledge”. Wefavor a “unified theory” that specifies a viewpoint about the general phenomenon, butwhich also distinguishes different types of knowledge.

Knowledge is a tested, evaluated and surviving structure of information (e.g. DNAinstructions, synaptic structures, beliefs, or claims) that may help the living systemthat developed it to adapt. This is our general viewpoint. It is consistent with ourdefinition of information. And it is consistent with CAS theory and the view thatknowledge is something produced by CASs in order to help them adapt toenvironmental challenges.

There are three types of knowledge:

(1) Tested, evaluated, and surviving structures of information in physical systemsthat may allow them to adapt to their environment (e.g. genetic and synapticknowledge).

(2) Tested, evaluated, and surviving beliefs (in minds) about the world (subjective,or non-sharable, mental knowledge).

(3) Tested, evaluated, and surviving, sharable (objective), linguistic formulationsabout the world (i.e. claims and meta-claims that are speech- or artifact-based orcultural knowledge).

The ontology reflected in the above definition is from Popper (1972, 1978, 1994, 1999;Popper and Eccles, 1977), but we have not used his terminology here. Figure 8illustrates the three types of knowledge and depicts their abstract relationships.

Figure 8.The three types ofknowledge

TLO12,2

198

Doing KM?At the beginning of this paper, we raised two questions related to the theme of thisspecial issue. Are KM practitioners “doing KM”, or are they “practicing” KM byhelping fields or techniques such as IT, CM, CRM, data warehousing, social networkanalysis, storytelling, CoPs, data mining, quality management, human resources, and“knowledge” cafes to “colonize” it? Is conceptual drift in KM so widespread that onecan conclude that, generally speaking, at least, KM as a formal, intentional endeavorhas, indeed, not yet been done? The detailed answers to these questions depend onone’s conceptual orientation to KM. Now that we have laid that orientation out, we canoffer an analysis that will provide some answers.

First, we do think that KM as a formal intentional endeavor has been done, and laterwe will discuss a case study that will illustrate this in detail. Having said that,however, we also believe that far too many “KM” efforts are not KM at all, butrepresent activities in fields peripheral to KM that “colonize” it by using KMterminology to mischaracterize non-KM interventions as instances of KM with theintention of benefiting from its cachet.

Second, in fact, such colonization of KM is not new. KM has been subject to it from thebeginnings of the discipline, when it was frequently characterized as being about“delivering the right information to the right people at the right time”, through use of theright IT tool. Thus, KM was viewed as an activity that encompassed deploying the rightIT tool in the enterprise and, often, using it to “manage knowledge” as characterizedabove. In that spirit, data warehousing, data mining, business intelligence (BI) and onlineanalytical processing (OLAP), business performance measurement (BPM), CRM,enterprise resource planning (ERP), collaboration management, groupware, search andretrieval applications, CM, semantic network/text mining applications, documentmanagement, image management, e-conference applications, e-learning applications,expertise locators (Yellow Pages), best practices database applications, and enterpriseinformation portals (EIPs), have all been characterized as KM tools, and projectsinvolving the deployment and use of one or another of these tools have beencharacterized and reported as KM projects. EIPs, in fact, were characterized as KM’s“killer app”, and scores of “KM cases” involving EIP projects were described andanalyzed in the KM and portal literature (Firestone, 2003b). At present, it iscommonplace for portal vendors to characterize their search and retrieval capabilities asKM capabilities, as if in using them one was automatically managing “knowledge” andalso finding it. Of all of the above IT applications, the most widely deployed in the firstgeneration of KM was the best practice database application.

Third, in our view, the association of the idea of “KM intervention” with any of theabove tools is frequently an instance of “conceptual drift”, mistaking KM for otherforms of activity. Such drift is harmful to KM because, ultimately, it confuses therecord of KM performance and therefore prevents an evaluation of KM based on thatperformance. Thus, because of conceptual drift it is possible to say that KM projectshave failed 85 percent of the time, when, in fact, neither KM interventions, nor anevaluation of them in any quantity, may actually have occurred.

However, how can one tell in any individual case of a project or program, that it is,in fact, a KM intervention, rather than an intervention of another type? The answer isthat one must evaluate an intervention:

Doingknowledge

management

199

. with one’s ideas about KM, knowledge processing, knowledge, information,other business processes and outcomes, and the differences among them in mind;and

. as if its classification as a KM intervention is conjectural and must be evaluatedagainst alternative conjectural classifications.

Of course, the quality of one’s evaluation will be dependent on the quality of one’s KMframework, and also on the extent to which one has considered other alternativeclassifications for the intervention.

Fourth, the situation with respect to analytical or social interventions is quiteanalogous to that of IT tools. As KM developed, CoPs (Wenger, 1998) became apopular, even dominant, “KM” intervention. Soon it was supplemented withstorytelling (Denning, 2001) interventions encouraging knowledge workers to usestories to both “sell” KM internally, share knowledge, and facilitate collaboration. Morerecently, social network analysis (SNA) (Cross and Parker, 2004) is being used todiscover the structure of relationships in existing communities, as well as the existenceof clusters of social relationships that can form the nuclei of new communities not yetself-organized. Another technique that has been popular is the knowledge cafe (Isaacs,1999), a technique in which participants circulate among multiple small interactivegroups carrying on a discussion of a selected topic and sharing their knowledge overthe course of a day. Additional techniques include “knowledge” auditing and mapping,value network analysis (Allee, 2003), Group decision-making processes, influencenetwork analysis, various quality management techniques, and, of course, culturalanalysis.

Our view of this list of techniques is analogous to our view about IT tools.Specifically, projects or programs that use them are not automatically, or even bypresumption, KM projects or programs. Whether they are, or not, depends on how thespecific intervention is related to KM, knowledge processing, information, knowledge,and so on, and also on how it is related to other management and knowledgeprocessing activities.

Fifth, “KM” interventions will involve either IT tools or social techniques or somemix of them. Whether such an intervention is a bona fide KM intervention depends onwhether it is a policy, program, or project targeted at enhancing knowledge processingand through knowledge processing, knowledge outcomes, and ultimately businessdecisions and processes. In other words it depends on whether, and on the extent towhich, the intervention fits the pattern expressed in the three-tier framework (seeFigure 1), and is targeted at the KLC (see Figure 6) as compared with the extent towhich it fits the pattern characteristic of other forms of management activity.

These considerations suggest that we apply the following criteria in deciding thequestion of whether an intervention is a KM intervention or something else:

(1) Is the intervention aimed at having an impact on problem recognition in DECsand business processes, on the KLC, or on some aspect of KM itself?

(2) If the intervention is aimed at some aspect of knowledge integration in the KLC,or the DOKB itself, does it incorporate a way of telling the difference betweenknowledge and information so that its impact is aimed at knowledge integrationand not just at information integration?

TLO12,2

200

(3) If the intervention is aimed at enhancing information acquisition relevant to aproblem, does it incorporate a way of telling whether external information is oris not relevant to the problem?

(4) If the intervention is aimed at enhancing knowledge claim formulation, does itincorporate tools or techniques for enabling creation of alternative knowledgeclaims?

(5) If the intervention is aimed at knowledge claim evaluation, does it incorporatetools or techniques that will enable testing and evaluation of knowledge claims?

(6) If the intervention is aimed at individual and group learning, does it meet any ofthe foregoing criteria about problem recognition, knowledge integration, or anyof the knowledge production sub-processes?

(7) If the intervention is aimed at enhancing KM itself, does it incorporate tools ortechniques that facilitate any of the following:. any aspect of producing or integrating KM-level knowledge;. problem recognition in KM-level DECs or business processes;. KM-level leadership;. building external relations with others in KM; (e) KM-level symbolic

representation;. changing knowledge processing rules;. crisis handling in KM;. negotiating for resources; and. allocating KM resources?

Sixth, while we do not have the space to apply these criteria to all of the techniques andtools listed above, we will apply them to a few of the most visible “KM” tools,techniques, and interventions. In the early days of KM, the most popular interventionwas the development of best practice database applications. The simple idea behindthis type of solution is that the quality of decisions will improve if “best practices” arecaptured, made available to knowledge workers, and reused by them. Are “bestpractices” interventions instances of “KM”? According to the criteria we have specifiedabove they are not, because while such systems certainly provide for sharingknowledge claims, they provide no way of differentiating knowledge from mereinformation, so one cannot tell whether knowledge or information is being sharedthrough them. In order for best practices systems to become KM interventions, theywould need to incorporate meta-claims describing the track record of performance or atleast the basis behind the best practice claims recorded in them. We have argued this atgreater length elsewhere (Firestone and McElroy, 2003a, Ch. 7).

Another popular “KM” intervention is the enterprise information/knowledge portal.One of us has distinguished these two types of portals sharply since early in 1999(Firestone, 1999). But as the terms are used by most in KM, whether an application iscalled one or the other seems to be unrelated to any systematic difference in theinterventions being discussed. In any event, portal tools and interventions, in spite ofthe early characterization of portals as “KM’s killer app” (Roberts-Witt, 1999), withperhaps a few exceptions for custom portal applications, fail to meet the above criteria

Doingknowledge

management

201

for KM interventions (Firestone, 2003b, Chs 13-17). Portals, like best practices systems,do not provide a way of distinguishing information from knowledge. As aconsequence, any support they provide for integration functions such as broadcasting,sharing, teaching (through e-learning applications), and search and retrieval, isrestricted to information, rather than knowledge, integration. Nor do portals generallyprovide targeted support for problem recognition, or for individual and group learning,or for knowledge claim evaluation. Nor do they provide targeted support for any of theKM activities distinguished in criterion 7 above.

There are the remaining possibilities that portal applications provide the requiredsupport for information acquisition and for knowledge claim formulation. But in thearea of information acquisition, portal applications have shortcomings in the extent towhich they support search results that are specifically relevant to problems. Althoughsearch technology has improved substantially since portals originated in 1998, it iswidely recognized that they do not provide results that are sufficiently targeted onproblems without a great deal of continuous interaction between humans and theportal. Moving to knowledge claim formulation, many portal interventions focused onCM or collaborative capabilities do not provide support for idea management, semanticnetworking, formal modeling, simulation, or other techniques supporting alternativeformulations. However, portals with strong structured data analysis/on-line analyticalprocessing/business intelligence capabilities support knowledge claim formulationincluding the specification of alternative claims. These types of portals supportknowledge processing and therefore interventions that deploy such portals are, indeed,KM interventions.

In brief, while portals provide a wide range of generalized support for informationprocessing and management, portals focused on content management provide littlespecific support for knowledge processing as outlined in the criteria mentioned earlier.It is not impossible for portals to provide support in many of these areas, and hence forKM interventions based on portals to enhance knowledge processing. All it requires isthat portal interventions incorporate portlets targeted at enhancing KLC functions.And, in fact, portals that support structured data analysis already provide support forknowledge claim formulation. But portal studies (Firestone, 2003b, Collins, 2003, Terraand Gordon, 2003) show that they mostly focus on CM, collaboration, documentmanagement, publication, CRM, imprecise searching, publication, taxonomydevelopment (a narrow type of knowledge production), and other forms oforganizational support, that are not directly related to knowledge processing. Thus,many portal interventions are not KM interventions, and to determine which ones arerequires analysis of the details of the portal application involved.

Turning to some examples from the area of social techniques for KM interventionswe have listed, we think it is also the case that CoPs, storytelling, and SNA-basedinterventions may or may not be KM interventions, depending on the details of thespecific intervention that is planned and implemented. Since CoP-based interventionsare among the favorite initiatives of knowledge managers, we begin by asking thequestion, when is a CoP intervention not a KM intervention? If the CoP intervention isaimed at enhancing knowledge sharing, but fails to provide a way of distinguishingCoP-produced knowledge from CoP-produced information, then, we claim, it is not aKM intervention but an information management (IM) intervention. How widespreadare such CoP-based interventions? While we have no data on this point, we believe that

TLO12,2

202

most CoP interventions are intended to enhance knowledge sharing but do not providea way of distinguishing knowledge from information, and therefore that most are notKM interventions at all.

KM interventions that attempt to introduce the use of storytelling as a technique ofknowledge sharing, share with CoP interventions the difficulty that they do not help todistinguish knowledge from information in what is shared. Stories are notautomatically knowledge because humans tell them.

On the other hand, they are, automatically, a way of expressing knowledge claims,so that interventions enhancing the capacity to express knowledge claims in thecontext or form of stories may be viewed as KM interventions, assuming that they alsoenhance the capability to express alternative knowledge claims. In addition,interventions that enhance the storytelling capabilities of knowledge managers maybe viewed as KM interventions, since they enhance both the leadership and knowledgeclaim formulation capabilities of knowledge managers.

A technique experiencing increasing popularity this year is SNA (Cross and Parker,2004), and one well-known KM blogger (Pollard, 2004) has even suggested that KM bere-invented as “social network enablement”, meaning that KM interventions would aimat enhancing opportunities for social networks to form and thrive. SNA is clearly ananalytic technique that can help generate knowledge claims about social networks, sointerventions whose aim is to provide IT tools for performing SNA, or training in SNA,are certainly narrow-scope KM interventions since they enhance knowledge claimformulation including generating alternative social network models. But socialnetwork enablement as a management intervention is aimed directly at enhancingsocial network formation and maintenance and not at any knowledge process per se.Therefore, it cannot be a KM intervention technique, generally speaking, except whenit is used to build KM-level external relationships, or as an aid in CoP or team-buildinginterventions that are aimed at enhancing KM processes or various knowledgesub-processes in the KLC.

We hope the foregoing discussion of best practices systems, EIPs, CoPs,storytelling, and social network enablement makes clear the following points. Mostinterventions that have been viewed as KM interventions have not been instances ofKM at all. Nor is it possible, in many instances, to conclude that an intervention is a KMintervention based on the tool or social technique it uses. As the old saying goes, thedevil is in the details, which, in turn, determine whether a particular intervention willfit one of the seven criteria we have specified earlier. In short, in many cases, whereothers think KM has been done frequently, our analysis implies that perhaps it has notbeen done. But having argued for that view, we now illustrate that KM both can be andhas been done. Our illustration is the Partners HealthCare case to which we now turn.

The Partners HealthCare caseIn July, 2002, authors Tom Davenport and John Glaser published a case study inHarvard Business Review (Davenport and Glaser, 2002) involving a KMimplementation at Partners HealthCare in Boston. Davenport is a KM researcherand consultant, and Glaser is the CIO at Partners.

The decision to invest in KM at Partners was largely driven by the cost of medicalerrors in healthcare, especially as reported by the Institute of Medicine (IOM) (Kohn et al.,1999) in 1998. According to IOM’s report, more than a million injuries and as many as

Doingknowledge

management

203

98,000 deaths each year are attributable to medical errors. At Partners, medical errors, asmeasured by them in 1995, showed that “more than 5 percent of patients had adversereactions to drugs while under medical care; 43 percent of those inpatient reactions wereserious, life threatening, or fatal. Of the reactions that were preventable, more than halfwere caused by inappropriate drug prescriptions” (Davenport and Glaser, 2002, p. 5).Moreover, “A study of the six most common laboratory tests ordered by physicians inBrigham and Women’s surgical intensive care unit found that almost half of the testsordered were clinically unnecessary” (Davenport and Glaser, 2002, p. 6).

On the basis of these and other problems discovered at Partners, the decision toinvest in KM was confined to the order-entry system, “because the system is central tophysicians delivering good medical care. When doctors order tests, medications, orother forms of treatment, they’re translating their judgments into actions. This is themoment when outside knowledge is most valuable. Without the system, doctors wouldhave no easy way to access others’ knowledge in real time” (Davenport and Glaser,2002, p. 6).

Perhaps the most informative part of the Davenport and Glaser case was theirdescription of how the KM system at Partners works. Here it is (Davenport and Glaser,2002, p. 7):

Here’s how it works. Let’s say Dr Goldzer has a patient, Mrs Johnson, and she has a seriousinfection. He decides to treat the infection with ampicillin. As he logs on to the computer toorder the drug, the system automatically checks her medical records for allergic reactions toany medications. She’s never taken that particular medication, but she once had an allergicreaction to penicillin, a drug chemically similar to ampicillin. The computer brings thatreaction to Goldzer’s attention and asks if he wants to continue with the order. He asks thesystem what the allergic reaction was. It could have been something relatively minor, like arash, or major, like going into shock. Mrs Johnson’s reaction was a rash. Goldzer decides tooverride the computer’s recommendation and prescribe the original medication, judging thatthe positive benefit from the prescription outweighs the negative effects of a relatively minorand treatable rash. The system lets him do that, but it requires him to give a reason foroverriding its recommendation.

Of central importance to the design of the integrated order-entry/KM system atPartners was the formation of centralized committees who were given theresponsibility to “create and maintain the knowledge repository” (Davenport andGlaser, 2002, p. 8) Only “clinicians at the top of their game” (Davenport and Glaser,2002, p. 8) were permitted to sit on these committees, and were given the authority “toidentify, refine, and update the knowledge used in each [medical/clinical] domain”. Itwas one of these committees of experts that was the source of the knowledge presentedto Dr Goldzer in the anecdote quoted above.

However, despite the authoritative source of the knowledge presented to physiciansat the time of order entry, Partners took a position of deference with respect to thedecisions made by front line, practicing physicians in the hospital. They reasoned:

With high-end knowledge workers like physicians it would be a mistake to remove themfrom the decision-making process; they might end up resenting or rejecting the system ifit changed their role – and with good reason. Because over-reliance on computerizedknowledge can easily lead to mistakes, Partners’ system presents physicians withrecommendations, not commands. The hope is that the physicians will combine theirown knowledge with the system’s (Davenport and Glaser, 2002, p. 8).

TLO12,2

204

As a result of the integrated order-entry/KM system at Partners, several benefits in theform of reduced medical errors were realized:

Out of the 13,000 orders entered on an average day by physicians at Brigham and Women’s,386 are changed as a result of a computer suggestion. When medication allergies or conflictwarnings are generated, a third to a half of the orders are canceled. The hospital’sevent-detection system generates more than 3,000 alerts per year; as a result of these alerts,treatments are changed 72 percent of the time – a sign that the hybrid human-computersystem at Partners is working as it should (Davenport and Glaser, 2002, pp. 8-9).

Also illustrative of the impact that KM had at Partners were the following results(Davenport and Glaser, 2002, p. 8):

. A controlled study of the system’s impact on medication errors found thatserious errors were reduced by 55 percent.

. When Partners experts established that a new drug was particularly beneficialfor heart problems, orders for that drug increased from 12 percent to 81 percent.

. When the system began recommending that a cancer drug be given fewer timesper day, the percent of orders entered for the lower frequency changed from 6percent to 75 percent.

. When the system began to remind physicians that patients requiring bed restalso needed the blood thinner heparin, the frequency of prescriptions for thatdrug increased from 24 percent to 54 percent.

From this case, Davenport and Glaser (2002, p. 6) concluded that the key to success inKM “is to bake specialized knowledge into the jobs of highly-skilled workers – to makethe knowledge so readily accessible that it can’t be avoided”. They further concludethat:

While there are several ways to bake knowledge into knowledge work, the most promisingapproach is to embed it into the technology that workers use to do their jobs. That ensuresthat knowledge management is no longer a separate activity requiring additional time andmotivation.

We believe that this method could revolutionize knowledge management in the same waythat just-in-time systems revolutionized inventory management – and by following much thesame philosophy (Davenport and Glaser, 2002, p. 6).

Analysis of the Partners HealthCare caseLet’s look at the Partners HealthCare case from the perspective of the conceptualframeworks used to evaluate other KM interventions. The three-tier frameworksuggests, first of all, that the intervention implementing the Partners system is a bonafide KM intervention, since its purposes appear to be to enhance: knowledgeintegration into DECs and DOKBs, problem recognition and error elimination in DECs,and knowledge production and the quality of knowledge in response to problems. Onthe other hand, the framework also suggests that the system is actually a knowledgeprocessing system, rather than a “KM” system, as it is described by Davenport andGlaser. The knowledge processing system operates at two levels: the level of theindividual doctor, and also the level of the organization.

Doingknowledge

management

205

The doctor’s levelOne of the purposes of the Partners’ system was to reduce errors by upgradingknowledge at the point where doctors make decisions to order tests, medications, orother forms of treatment. Knowledge at the point of decision was to be upgraded byway of the new system’s ability to broadcast others’ knowledge to the decision maker,and also by the decision maker thereby using, or not, the shared knowledge to questionhis/her own decisions or actions. In other words, from the point of view of ourframeworks, the system is, in the first instance, about eliminating or reducing errors inDECs by increasing the frequency with which doctors question, critically evaluate, andrecognize problems in the decisions they are contemplating. The system is supposed tomake doctors look for problems in their views, and if they find them, initiate problemsolving (that is, KLCs) of their own, in the expectation that this will increase the qualityof the beliefs that survive and inform their order entry decisions.

Thus, in terms of the DEC framework, when Dr Goldzer uses his previousknowledge to decide to treat Mrs Johnson’s infection with ampicillin, he acts on thedecision by ordering the drug. The system prepares to intervene in Dr Goldzer’s DECbetween his action and the production of a result for him to monitor and evaluate.There were two options for the system in this situation. If it had not found anycontra-indicating history (or other previous knowledge) related to Goldzer’s order, hisorder would have been processed, and the results of Goldzer’s DEC would have beenthe administration of ampicillin to Mrs Johnson and its downstream effects. The optionapplicable to Goldzer’s actual situation, however, was that the knowledge claims in thesystem conflicted with his order, so the system intervened in Goldzer’s DEC andbrought Mrs Johnson’s previous allergic reaction to his attention by presenting himwith a knowledge claim about that as the result of his decision. Thus, it integrated theorganization’s knowledge into his DEC, and forced him to evaluate critically his beliefthat the right thing to prescribe for Mrs Johnson was ampicillin, against the knowledgeclaims it presented to him. In doing that, the system facilitated the possibility, or, if youlike, increased the probability, that Dr Goldzer would question his decision, recognize aproblem with it, and then initiate a knowledge life cycle to solve it.

In fact, that is what he did, initiate a KLC, and specifically, from his individualperspective, an activity of information acquisition. When Dr Goldzer learns that MrsJohnson’s allergic reaction to penicillin was a rash, he uses that information and hisjudgment that “the positive benefit from the prescription outweighs the negative effectsof a relatively minor and treatable rash”, to falsify the computer’s recommendation, theorganization’s knowledge, that he not prescribe ampicillin. He then ends his individualKLC and returns to the associated operational DEC, through which he again places hisoriginal order. Before he is allowed to proceed, however, he is prompted by the system tointegrate into the organizational DOKB knowledge claims and meta-claims explainingwhy he falsified (over-rode) the system’s knowledge claims.

The organizational perspectiveWhen we look at the order entry system from an organizational perspective, we seeknowledge production being performed by committees of experts. They evaluate whatgoes into the system, and the claims they approve receive the “imprimatur” of theorganization as knowledge to be integrated into order entry DECs when triggered byspecific transactions. In terms of Dr Goldzer’s activities, one of the committees of

TLO12,2

206

experts was the source of the order-relevant knowledge presented to him in thedescription quoted above.

From the viewpoint of our frameworks, the committees are continuously processingknowledge claims in an effort to reduce or eliminate the errors in the DOKB, and thusto upgrade its quality over time. The committees are designated authorities forknowledge production and knowledge claim evaluation at the organizational level,directed at solving the problem of medical error reduction in order entry. They performKLCs, evaluate, and select the knowledge claims that are formally designated asorganizational knowledge, and that will be made available through the system forintegration into the order entry DECs.

The system, however, works in such a way that the centralization of knowledgeproduction in the committee is balanced by the participation of all physicians inknowledge claim formulation and evaluation in the context of their participation in theorder entry system. Partners understood the need to maintain a distributed decisionmaking system with respect to order entry, and, in addition, to reinforce a distributedproblem solving system with respect to problems arising out of the order entrydecision. Partners did this because it recognized the fallibility of organizationalknowledge produced by the committees, the need to involve the doctors and theirknowledge in solving problems and adding knowledge claims to the DOKB, and theneed to view system interventions in decisions made by the doctors, as acts ofknowledge integration, intended to strengthen monitoring and evaluation and problemrecognition in the DEC, rather than knowledge imposition.

In the end, the Partners’ system is stronger because it is a distributedproblem-solving system, in which the committees, through the system, help thedoctors to recognize that there are problems with some of their orders. However, bysometimes insisting on their decisions and giving the committees feedback on theirown reasons for doing so, the doctors, again through the system, are providingknowledge claims to the committees, as well as critical evaluations of the committees’recommendations (i.e. their knowledge claims) to them, in the form of the reasons theyprovide for over-riding such recommendations.

When the committees later review the doctors’ claims, and decide whether toincorporate them into the system and/or modify their own previous recommendations,they are engaging in further knowledge claim evaluation and in producing newknowledge at the organizational level. In this way, the system links the individual levelwith the organizational level and makes the doctors participants, along with thecommittees, in organizational problem solving and knowledge production. In thisregard, the Partners’ system not only injects organizational knowledge at key decisionpoints in the order entry process, but also integrates knowledge processingfunctionality at the same time and place in the form of knowledge claim formulationand knowledge claim evaluation. On the one hand, the system broadcastsorganizational knowledge to the physicians and supports further informationretrieval as well, while on the other it engages them in various aspects oforganizational knowledge processing. Of most significance here is the requirement thatthe physician, Dr Goldzer, record his reasons for over-riding the expert committee’srecommendations. There is a dialectical dimension to the system. This suggests a newmetaphor or heuristic for KM. It is not just push versus pull anymore, it is push or pulland pull back!

Doingknowledge

management

207

Finally, Davenport and Glaser, in their account, characterize the Partners’ system asembedding KM into the business process, or “baking specialized knowledge intoknowledge work”. From the viewpoint of our frameworks, however, KM remainswhere it is, in the top-tier of the framework. The Partners’ system supports knowledgeprocessing at both individual and organizational levels. Organizational knowledge isintegrated into DECs so that problems are surfaced. More KLCs occur, evaluating theknowledge of both doctors and the organization, and, lastly, errors are more likely to bereduced or eliminated.

Implications for KM strategy and KM programsThe Partners HealthCare case is a great illustration of how to go about a successful KMintervention that enhances knowledge processing at the levels of both the individualand the organization in such a way that the changes have an impact on businessoutcomes: in this case, lives saved and serious consequences of medical errors avoided.The case also leads us to suggest an extension of the pattern into a KM strategy that,we propose, is at once coherent and incremental.

The vision of the strategy is to enhance knowledge processing in the enterprisegradually in a manner that will add increasing value and create sustainable innovationover time. The end state of the strategy is attaining a form of organization called theOpen Enterprise, which, theory suggests, is an environment providing maximalsupport for sustainable innovation, problem solving, and adaptation. The OpenEnterprise, an extension of Karl Popper’s ideas about the Open Society (Popper, 1945)to organizations, is open to:

. New problems recognized by any of its agents.

. New ideas generated by any of its agents (knowledge claim formulation).

. Continuous criticism of previously generated ideas by any of its agents(knowledge claim evaluation).

The Open Enterprise is not democratic in decision making, nor in management, but itrequires at least internal transparency and inclusiveness in distributed knowledgeprocessing and problem-solving. Here are the steps in the strategy:

(1) Use a formal KM methodology to implement the strategy. There are manymethodologies available that apply to KM tools and techniques, but there isvery little in the literature offering a comprehensive KM program and projectmethodology. Perhaps the only alternative is K-STREAMe, a recentformulation of our own (Firestone and McElroy, 2004b, c, KMCI, 2004).

(2) Identify and prioritize decisions (DECs), work flows, or business processesaccording to risk. In formulating a KM strategy and an associated program, oneneeds systematically to specify DECs and, where necessary, work flows, orbusiness processes that can produce highly negative business outcomes iferrors are made. In the Partners case, the organization identified a decision, theorder entry decision, involving high risk for the organization. That decision wasthe source of costly medical errors, and having a favorable impact on it waslikely to produce a lot of social credit for the KM function at Partners.Identification of high risk DECs should be followed by prioritization of themaccording to risk, taking into account, ease and expense of intervention, and

TLO12,2

208

likelihood of success. It should also be understood that whereas high-riskdecisions are the logical place to start in this approach, decision-oriented KMinterventions do not end there. All decisions in organizations are subject toenhancement, and the opportunity to improve performance in more generalterms exists across the board. Demonstrating success for high-risk decisionsfirst, however, is appropriate, both for purposes of having impact early in KMprograms and also for building confidence and justification for furtherinvestment in KM.

(3) Select DECs, work flows, or business processes as targets for KM interventionsaccording to priority and develop the business case. There is no indication thatPartners selected their KM intervention in the order entry decision from a set ofalternatives. But an essential step in developing a long-term strategy is to usethe outcome of step 1 to perform such a selection, and in doing so to develop theoutlines of a KM program designed to enhance knowledge processing.

(4) If you can, make interventions that embed new knowledge processingfunctionality within existing IT-based business applications supporting DECs,work flows or business processes. This follows the pattern of the Partners case. Itassumes that you can find high risk DECs already supported by existing ITapplications to use as the objects of intervention. Intervening in existingapplications is preferable to introducing entirely new applications, becausepeople already depend on these applications as part of their job, and are likely tocontinue to use the enhanced system. Note here, however, that the integration ofknowledge processing functions at key decision points need not necessarilytake the form of IT implementations. Process or procedural changes can also bemade, the effects of which will cause key decisions, and the knowledge claimsbehind them, to be tested by others before being put into action.

(5) Make sure the new functionality added to the IT business application, or process,presents competing organizational knowledge claims to those expressed orimplied in a DEC outcome. This is needed to encourage questioning of previousindividual-level knowledge in DECs, which, in turn, can encourage increasedproblem recognition, individual KLCs, and error reduction in key decisions.Knowledge processing systems resulting from such KM interventions will helpdecision makers to “look for trouble,” recognize problems, and initiate KLCs,and in the process will bring inclusiveness to problem recognition and problemsolving related to the high risk area which is the target of the intervention.

(6) When competing knowledge claims introduced in knowledge integration areover-ridden by a decision maker, new IT application (or process) functionalityshould require that the superseding knowledge claims and meta-claims be addedto the DOKB by the decision maker. This is essential to accumulate a trackrecord of knowledge claim performance in the DOKB. Soliciting knowledgeclaims and meta-claims in this way opens up knowledge processing to newideas and to distributed knowledge claim evaluation. Thus, it moves theorganization closer to the Open Enterprise by including the decision maker(knowledge worker) in knowledge processing, and in knowledge productionspecifically.

Doingknowledge

management

209

(7) Once the first intervention is completed, continue implementing the KMprogram, project by project, according to the priority established earlier.Following the strategy will strengthen the ability to: recognize problems in areaafter area, initiate KLCs, produce distributed problem solving, and increaseadaptiveness. And in the process it will move the organization closer to theOpen Enterprise, problem area by problem area, through creating transparency,inclusiveness and other characteristics of the Open Enterprise in each case.That progress should be tracked and measured, since the closer theorganization gets to the Open Enterprise, the more it will exhibitadaptiveness and sustainable innovation.

Summary and conclusionsKM as a field has been characterized by a great deal of confusion about its conceptualfoundations and scope. As a result, practitioners have tended to view KM interventionsas those that have been given that name by themselves or others who claim to bepractitioners. In this paper, we have suggested that continuing that practice isdestructive to KM as a discipline, because it prevents coherent evaluations of KM’strack record. Moreover we have offered a framework and set of criteria based on it fordeciding whether claimed interventions are bona fide instances of KM, and illustratedthe use of that framework in critical evaluation of typical “KM” interventions,including extensive discussion of an unambiguous case where KM has been done.

This case, the well-known Partners HealthCare project, was also shown to illustratea pattern of intervention that can serve as the basis of a long-term systematic strategyfor implementing KM in the enterprise. The strategy is risk-based. It is one that candeliver concrete, incremental solutions and benefits to the enterprise by creatingquality-control systems for knowledge-in-use as a support for distributed decisionmaking and knowledge processing. In the long run, it can transform the enterprise intoan organizational form that we call the Open Enterprise, and thereby supportsustainable innovation and help solve the general problem of organizationaladaptiveness and performance.

References

Ackoff, R. (1970), A Concept of Corporate Planning, Wiley-Interscience, New York, NY.

Allee, V. (2003), The Future of Knowledge, Butterworth-Heinemann, Burlington, MA.

Argyris, C. (1993), Knowledge for Action, Jossey-Bass, San Francisco, CA.

Argyris, C. and Schon, D. (1974), Theory in Practice: Increasing Professional Effectiveness,Jossey-Bass, San Francisco, CA.

Bickhard, M. (1999), “Information and representation in autonomous agents”, available at:www.lehigh.edu/,mhb0/infrepautagents.html

Bickhard, M. (2000), “The dynamic emergence of representation”, available at: www.lehigh.edu/,mhb0/dynemergrep.html

Campbell, D. (1974), “‘Downward causation’ in hierarchically organized biological systems”,in Ayala, F. and Dobzhansky, T. (Eds), Studies in the Philosophy of Biology, Macmillan,London.

TLO12,2

210

Cavaleri, S. and Reed, F. (2000), “Designing knowledge-generating processes”, Knowledge andInnovation: Journal of the KMCI, Vol. 1 No. 1, pp. 109-31.

Cavaleri, S. and Reed, F. (2001), “Organizational inquiry: the search for effective knowledge”,Knowledge and Innovation: Journal of the KMCI, Vol. 1 No. 3, pp. 27-54.

Collins, H. (2003), Enterprise Knowledge Portals, AMACOM, New York, NY.

Cross, R. and Parker, A. (2004), The Hidden Power of Social Networks, HBSP, Boston, MA.

Davenport, T. and Glaser, J. (2002), “Just-in-time delivery comes to knowledge management”,Harvard Business Review, July, pp. 107-11.

Denning, S. (2001), The Springboard, KMCI Press/Butterworth-Heinemann, Boston, MA.

Firestone, J.M. (1999), Enterprise Information Portals and Enterprise Knowledge Portals, DKMSBrief, No. 8, Vol. 20, Executive Information Systems, Wilmington, DE, March 20, availableat www.dkms.com/White_Papers.htm

Firestone, J.M. (2000a), Knowledge Management: A Framework for Analysis and Measurement,White Paper, No. 17, October 1, available at: www.dkms.com/White_Papers.htm,Executive Information Systems, Wilmington, DE.

Firestone, J.M. (2000b), “Enterprise knowledge portals: what they are and what they do”,Knowledge and Innovation: Journal of the KMCI, Vol. 1 No. 1, pp. 85-108, available at:http://www.dkms.com/White_Papers.htm

Firestone, J.M. (2003a), “Minding the (knowledge) gap”, Knowledge Management Magazine, Vol. 6No. 8, April, pp. 20-4.

Firestone, J. (2003b), Enterprise Information Portals and Knowledge Management, KMCIPress/Butterworth-Heinemann, Burlington, MA.

Firestone, J.M. (2003c), How Knowledge Management Can Help Identify and Bridge KnowledgeGaps: An EIS Professional Paper, Executive Information Systems, Wilmington, DE,available at: www.dkms.com/professionalpapers.htm

Firestone, J. and McElroy, M. (2003a), Key Issues in the New Knowledge Management, KMCIPress/Butterworth-Heinemann, Burlington, MA.

Firestone, J. and McElroy, M. (2003b), “The new knowledge management”, KnowledgeManagement Magazine, Vol. 6 No. 10, June, pp. 12-16.

Firestone, J. and McElroy, M. (2003c), Excerpt #1 from The Open Enterprise: Building BusinessArchitectures for Openness and Sustainable Innovation, KMCI Online Press, Hartland FourCorners, VT, available at: www.dkms/com, www.macroinnovation.com and www.kmci.org

Firestone, J. and McElroy, M. (2004a), “Viewpoint: organizational learning and knowledgemanagement: the relationship”, The Learning Organization, Vol. 11 No. 2, pp. 177-84.

Firestone, J. and McElroy, M. (2004b), “K-STREAMe and the new knowledge management?”,All Life Is Problem Solving, May 12, available at: http://radio.weblogs.com/0135950/

Firestone, J. and McElroy, M. (2004c), “Are there core tools and techniques of knowledgemanagement?”, All Life is Problem Solving, May 12, available at: http://radio.weblogs.com/0135950/

Gell-Mann, M. (1994), The Quark and the Jaguar, W.H. Freeman, New York, NY.

Haeckel, S.H. (1999), Adaptive Enterprise, Harvard Business School Press, Boston, MA.

Hall, W. (2005), “The biological nature of knowledge in learning organizations”, The LearningOrganization, Vol. 12 No. 2.

Holland, J.H. (1995), Hidden Order, Addison-Wesley, Reading, MA.

Holland, J.H. (1998), Emergence, Addison-Wesley, Reading, MA.

Doingknowledge

management

211

Isaacs, D. (1999), “Knowledge cafe presentation”, paper presented at the Enterprise IntelligenceConference, Lake Buena Vista, FL, December 7.

Juarrero, A. (1999), Dynamics in Action, MIT Press, Cambridge, MA.

Kauffman, S. (1995), At Home in the Universe, Oxford University Press, New York, NY.

KMCI (2004), “CKIM certificate program”, available at: www.kmci.org/ckim-certificate.html

Kohn, L., Corrigan, J. and Donaldson, M. (Eds) (1999), To Err Is Human: Building a Safer HealthSystem, Committee on Quality of Health Care in America, Institute of Medicine, NationalAcademy Press, Washington, DC.

Kolb, D. (1984), Experiential Learning, Prentice-Hall, Englewood Cliffs, NJ.

Kolb, D. and Fry, R. (1975), “Toward an applied theory of experiential learning”, in Cooper, C.(Ed.), Theories of Group Process, John Wiley & Sons, London.

McElroy, M.W. (1999), “The second generation of KM”, Knowledge Management, October,pp. 86-8.

McElroy, M.W. (2000), “The new knowledge management”, Knowledge and Innovation: Journalof the KMCI, Vol. 1 No. 1, October 15, pp. 43-67.

McElroy, M.W. (2003), The New Knowledge Management: Complexity, Learning, and SustainableInnovation, KMCI Press/Butterworth-Heinemann, Burlington, MA.

Pollard, D. (2004), “Social networking, social software and the future of knowledge management”,How to Save the World, available at: http://blogs.salon.com/0002007/2003/05/28.html#a251

Popper, K.R. (1945), The Open Society and its Enemies, Routledge & Sons, London.

Popper, K.R. (1972), Objective Knowledge, Oxford University Press, London.

Popper, K.R. (1978), “Three worlds: the Tanner Lecture on Human Values”, available at:www.tannerlectures.utah.edu/lectures/popper80.pdf

Popper, K.R. (1987), “Natural selection and the emergence of mind”, in Radnitzky, G. and Bartley,W.W. III (Eds), Evolutionary Epistemology, Rationality, and the Sociology of Knowledge,Open Court, La Salle, IL.

Popper, K.R. (1994) in Notturno, M.A. (Ed.), Knowledge and the Body-Mind Problem, Routledge,London.

Popper, K.R. (1999), All Life Is Problem Solving, Routledge, London.

Popper, K.R. and Eccles, J.C. (1977), The Self and Its Brain, Springer, Berlin.

Roberts-Witt, S.L. (1999), “Making sense of portal pandemonium”, KM Magazine, July, pp. 37-53.

Senge, P. (1990), The Fifth Discipline, Currency Doubleday, New York, NY.

Terra, J. and Gordon, C. (2003), Realizing the Promise of Corporate Portals, KMCI Press/Butterworth-Heinemann, Burlington, MA.

Wenger, E. (1998), Communities of Practice: Learning, Meaning, and Identity, CambridgeUniversity Press, New York, NY.

TLO12,2

212